1. Joint Modeling
1.1 Trace plots for convergence check
The current MCMC setting is:
- 4000000 iteration;
- 3000000 burn-in;
- 1000 thinning.
1.2 Gelman and Rubin’s convergence check
## Potential scale reduction factors:
##
## Point est. Upper C.I.
## HDevsum 1.001 1.006
## LDevsum 0.999 0.999
## dh0 1.002 1.011
## dl0 1.005 1.006
## dl1 1.000 1.004
## dl2 0.999 1.000
## dl3 1.000 1.003
##
## Multivariate psrf
##
## 1
1.3 ACF Plots
Here we plotted ACF plots for the following variables:
- Total deviance;
- Variables that didn’t pass the convergence check.
1.4 WAIC results
| LevelH | LevelL | |
|---|---|---|
| DIC | 1358.44190 | 24731.751 |
| DIC3 | 1238.42518 | 21023.763 |
| PWAIC | 80.44013 | 1266.063 |
| WAIC | 1296.83208 | 21589.610 |
2. Separate Modeling of High-Level
2.1 Trace plots for convergence check
The current MCMC setting is:
- 4000000 iteration;
- 3000000 burn-in;
- 1000 thinning.
2.2 Gelman and Rubin’s convergence check
## Potential scale reduction factors:
##
## Point est. Upper C.I.
## HDevsum 1.00 1.02
## dh0 1.01 1.03
##
## Multivariate psrf
##
## 1.01
2.3 ACF Plots
Here we plotted ACF plots for the following variables:
- Total deviance;
- Variables that didn’t pass the convergence check.
2.4 WAIC results
| H0b | |
|---|---|
| DIC | 1396.91005 |
| DIC3 | 1260.67413 |
| PWAIC | 94.92954 |
| WAIC | 1333.02177 |
3. Separate Modeling for Low-level
3.1 Trace plots for convergence check
The current MCMC setting is:
- 4000000 iteration;
- 3000000 burn-in;
- 1000 thinning.
3.2 Gelman and Rubin’s convergence check
## Potential scale reduction factors:
##
## Point est. Upper C.I.
## LDevsum 0.999 0.999
## dl0 1.031 1.139
## dl1 1.002 1.012
## dl2 1.017 1.081
## dl3 0.999 0.999
##
## Multivariate psrf
##
## 1.03
3.3 ACF Plots
Here we plotted ACF plots for the following variables:
- Total deviance;
- Variables that didn’t pass the convergence check.
3.4 WAIC results
| L7 | |
|---|---|
| DIC | 24158.821 |
| DIC3 | 20960.738 |
| PWAIC | 1182.775 |
| WAIC | 21473.510 |